Newly emerging IADL impairment predicts subsequent cognitive decline across four population-based aging cohorts

Scalable clinical markers of neurocognitive vulnerability in aging are needed, particularly in settings where biomarker access remains limited. Instrumental activities of daily living (IADLs) capture cognitively demanding everyday behavior and may provide a low-cost, repeatable functional marker of elevated risk. We harmonized longitudinal data from four population-based aging cohorts-CHARLS, ELSA, HRS, and SHARE-including adults aged ≥50 years with eligible baseline assessments. IADL impairment was defined as difficulty in at least one instrumental activity, and global cognition was harmonized within cohort as a standardized cognitive z-score. We first examined whether baseline IADL impairment was associated with subsequent cognitive slope, and then tested whether prior-wave IADL impairment predicted subsequent annualized cognitive change using lagged longitudinal models restricted to participants without baseline IADL impairment. In this context, newly emerging IADL impairment refers to IADL difficulty observed during follow-up among participants who were unimpaired at baseline. Dose-response analyses evaluated the number of impaired IADL domains. Cohort-specific estimates were generated separately and pooled using random-effects meta-analysis. Prior-wave IADL impairment predicted faster subsequent annualized cognitive decline (pooled β = -0.111 SD/year, 95% CI -0.141 to -0.081; p < 0.001). This corresponds to an expected difference of approximately -0.22 SD over 2 years, -0.44 SD over 4 years, and -0.67 SD over 6 years, assuming a constant linear difference in slope. A graded dose-response pattern was observed: compared with 0 impaired domains, 1 impaired domain was associated with β = -0.095 SD/year, ≥2 impaired domains with β = -0.147 SD/year, and each additional impaired domain with β = -0.063 SD/year. By contrast, baseline IADL impairment was not significantly associated with subsequent cognitive slope (β = 0.005 SD/year, 95% CI -0.002 to 0.011; p = 0.145). Findings were robust across sensitivity analyses and directionally consistent across cohorts, although effect magnitudes were heterogeneous. These findings support repeated IADL assessment as a pragmatic trigger for risk stratification and timely cognitive evaluation in initially functionally intact aging populations, rather than as definitive evidence of disease-specific neurodegenerative decline.